A real-time system for dynamic optical tomography

Share Embed


Descrição do Produto

A real-time system for dynamic optical tomography Christoph H. Schmitz, Harry L. Graber, Randall L. Barbour Dept. of Pathology, SUNY Downstate Medical Center, Box 25, 450 Clarkson Ave., Brooklyn, NY 11203, U.S.A. Tel.: 1 (718) 270-2832, Fax: 1 (718) 270-3313, Email: [email protected]

Joseph M. Lasker, Andreas H. Hielscher Depts. of Biomedical Engineering and Radiology, Columbia University, 530 West 120th St, NY, NY 10027

Yaling Pei NIRx Medical Technologies Corp., 15 Cherry Lane, Glen Head, NY 11545, U.S.A.

Abstract: Presented are the operating characteristics of an integrated CW-near infrared tomographic

imaging system capable of fast data collection and producing 2D/3D images of optical contrast features that exhibit dynamic behavior in tissue and other highly scattering media in real time. Results of preliminary in vivo studies on healthy and cancerous breast tissue are shown. OCIS codes: (170.6960) Tomography; (170.5280) Photon migration

1. Introduction Dynamic near-infrared optical tomography (DYNOT) is a new general-purpose noninvasive imaging tool capable of investigating functional states of the vasculature and its interaction with surrounding tissue structures [1,2]. In the current report, we present a description of an integrated hardware and software approach that provides for the imaging of dynamic processes in real-time. By way of example we show results obtained from a subject diagnosed with Stage II breast cancer that document altered functional states associated with the tumor vasculature. 2. Methods 2.1 Instrumentation The concept, design rationale, and implementation of a dynamic near-infrared optical tomographic (DYNOT) imaging system have been described in recent reports [3-5]. Here we briefly summarize system functionalities.

Data Analysis

System Control

Hardware

The system currently being tested operates at two wavelengths (up to four wavelengths are possible) and features a data acquisition rate of up to ~ 2.7 full tomographic data sets per second per wavelength, (equivalent to ~ 4,300 single measurements per second). This is achieved by employing a hybrid approach of time multiplexing the source locations and frequency encoding the illuminating wavelengths. The latter allows for the parallel readout of all up to 32 detector channels at different wavelengths simultaneously. Additionally, the system employs an optical detection scheme that has an exceptionally large dynamic range (180 dB). This permits examination of large tissue volumes while maximizing the achievable data acquisition rate. The large dynamic range is achieved by an electronic gain measuring Detection unit λ1 heads switching scheme that automatically λ2 3D (head, data D-MUX Σ breast, ...) storage adjusts the sensitivity range of each λ3 2D (limb, detector channel in sync with the abdomen, ...) λ4 changing source location on the target. Other functionalities include timing System timing hardware and a software protocol to CPU 1 Interactive control provide communication between the system components in order to orchestrated the necessary system tasks. This includes the advancement of the source positions, the timely adjustment of Image reconstruction the detector gain settings, and the CPU 2 triggering of the readout process. Time series analysis Another feature of the DYNOT system is the availability of exchangeable sensor heads having various geometries that form Fig. 1. Schematic of functionality and system architecture. λ1-4: lasers of different wavelengths, Σ: Overlaying beams, D-MUX: optical demultiplexer. optical interfaces suitable for a wide range

of applications. Available are sensors for the limbs (circular geometry), the breast (folding hemisphere), a helmetlike device for the head, and various two-dimensional fiber array designs for nearly planar geometries. 2.2. System operation and functionality Fig. 1 shows a schematic of the system’s architecture. The first level represents the system hardware. Indicated are the basic components that allow for the system’s performance capabilities described above. The second level comprises mainly the user interface that allows for system setup and real-time viewing of acquired measurements in various formats. These are configured within a LabVIEW environment. Various display modes are available to the user, including the selection of wavelength and the display of estimated changes in blood oxygenation and volume. A third level of functionality is controlled by a second CPU, operating under a Unix environment. This provides additional levels of data processing associated with image reconstruction, image display and image analysis. Various coupled forward-inverse reconstruction algorithms are available, which allow for image recovery using first-order or recursive finite element based schemes for 2D or 3D problems. All are derived from the diffusion equation for DC illumination. Computed parameters include absorption only, scattering (diffusion) only, or both. Schemes for real-time image recovery in the case of first order solutions are described in an accompanying report by Pei et al. The computed image series is available for off line interrogation using a spectrum of analysis routines embedded within an interactive MATLAB environment. These allow for the computation and display of linear and nonlinear properties associated with the time-varying pixel data. Results

Relative Hb / HbO2

Fig. 2 shows results obtained with the DYNOT system on 0.20 a patient diagnosed with Stage II breast cancer. Shown 1 2 3 1 2 3 are temporal changes in normalized oxy- and 0.15 deoxyhemoglobin levels as estimated from a dualwavelength measurement (760 nm and 830 nm) for a 0.10 selected source-detector pair near the tumor location on the right breast and for the corresponding source-detector 0.05 pair on the left (healthy breast). The indicated numbers refer to various breathing maneuvers the volunteer was 0.00 asked to perform (1 = deep breathing, 2 = breath hold, 3 = recovery). Comparison between the graphs reveals a qualitatively -0.05 different trend in hemoglobin states during a breath hold. In both graphs, the oxy- and deoxy-hemoglobin (HbO2, -0.10 400 650 900 500 750 1000 Hb, respectively) levels rise initially upon a breath hold, Time point which is expected because of the rise in venous return pressure will cause an increase in tissue blood volume. Fig. 2. Time course of hemoglobin states during a breath hold. Right: Following this we observe that contrary to the trend seen left (healthy) breast, left: right breast (w/ tumor). in the healthy breast, the HbO2 level in the tumor bearing breast becomes unstable and then declines accompanied by a steeper rise in the levels of deoxyhemoglobin. These Correlation Value results are entirely consistent with the well-recognized -1 E-8 general finding that solid tumors function on the brink of -7.5 E-9 hypoxemia. -5 E-9 In Fig. 3 we show an example of how this -2.5 E-9 information, extracted from continuous optical tomographic measurements, can serve to image tumor tissue with high contrast. Shown is an isocontour plot of the product of the time correlation values of the HbO2 and Fig. 3. Right: 3D map of correlation values for the temporal variations Hb levels seen during the breath-hold computed from the of oxygenation states in the tumor bearing breast. Left: Sonogram of 3D time-series image data. For presentation clarity, we the same breast, approx. at same scale. show only those contour levels that comprise the highest 90% of the computed values (i.e., background contrast is
Lihat lebih banyak...

Comentários

Copyright © 2017 DADOSPDF Inc.